Sensitive Detection of Multiple Rotavirus Genotypes with a Single ...

JOURNAL OF CLINICAL MICROBIOLOGY, Aug. 2008, p. 2547–2554 0095-1137/08/$08.00⫹0 doi:10.1128/JCM.02428-07 Copyright © 2008, American Society for Microbiology. All Rights Reserved.

Vol. 46, No. 8

Sensitive Detection of Multiple Rotavirus Genotypes with a Single Reverse Transcription–Real-Time Quantitative PCR Assay䌤 Ion Gutie´rrez-Aguirre,1* Andrej Steyer,2 Jana Boben,1 Kristina Gruden,1 Mateja Poljsˇak-Prijatelj,2 and Maja Ravnikar1 Department of Biotechnology and Systems Biology, National Institute of Biology, Ljubljana, Slovenia,1 and Institute of Microbiology and Immunology, Faculty of Medicine, University of Ljubljana, Ljubljana, Slovenia2 Received 18 December 2007/Returned for modification 20 December 2007/Accepted 28 May 2008

Rotaviruses are one of the major causes of diarrhea in infants and children under 5 years old, especially affecting developing countries. In natural disasters, fecal matter and potable waters can mix, allowing low, yet infective, concentrations of rotavirus to be present in water supplies, constituting a risk for the population. Any of the most commonly detected rotavirus genotypes could originate an outbreak. The development of a fast and sensitive method that could detect the broadest possible range of rotavirus genotypes would help with efficient diagnosis and prevention. We have designed a reverse transcription (RT)–real-time quantitative PCR approach targeted to the rotaviral VP2 gene, based on a multiple-sequence alignment of different human rotaviral strains. To overcome the high nucleotide sequence diversity, multiple forward and reverse primers were used, in addition to a degenerate probe. The performance of the assay was tested on isolates representing the most prevalent human genotypes: G1P[8], G2P[4], G3P[8], G4P[8], G9P[8], and G12P[8]. The developed method improved classical rotavirus detection by enzyme-linked immunosorbent assay and nested RT-PCR by 5 and at least 1 order of magnitude, respectively. A survey of 159 stool samples indicated that the method can efficiently detect a broad range of rotavirus strains, including different G-P genotype combinations of human, porcine, and bovine origin. No cross-reactivity was observed with other enteric viruses, such as astrovirus, sapovirus, and norovirus. (22). Animals have been postulated as a potential source of newly emerging rotavirus genotypes in humans (6). The most frequently used methods for the detection of rotaviruses are enzyme immunoassay, electron microscopy (EM), and conventional reverse transcription (RT)-PCR (8). A few RT–real-time quantitative PCR (qPCR) assays, which are based mainly on Sybr green chemistry, have been published (14, 17, 23). Only two rotavirus-targeted TaqMan RT-qPCR approaches have been developed, which were tested uniquely on human strains and proved to detect only type G1 and types G1, G2, and G4, respectively (15, 19). Rotaviruses are very stable in the environment, and they can remain infectious for weeks. This is the basis for the high incidence of rotavirus gastroenteritis, especially in winter months, when their stability, due to relative humidity and temperature conditions, is further increased (1, 6). It has been shown that raw food, untreated water, treated water, and irrigation water can represent possible sources of rotaviral gastroenteritis outbreaks (6, 30). As an example, in a recent study performed in southern Africa, the presence in irrigation waters of multiple rotavirus types (G1, G2, G8, and G9) was confirmed by using type-specific RT-PCR (30). In addition, swine or cattle slurry deposited in fields could contain animal rotavirus types, which could pass via runoff water into freshwater and later be transmitted to humans (6), causing illnesses of various severities (16). The potential presence of rotavirus genotypes of both animal and human origin in environmental waters (6, 30) and the proven ability of this virus for cross-species transmission (16) call for the development of a detection method with the potential to detect multiple rotaviral genotypes in order to apply

Group A rotaviruses constitute one of the most important causes of acute diarrhea in young children worldwide and are important pathogens in some animals, as well (8). In children under 5 years of age, rotaviruses are estimated to cause about 500,000 to 600,000 deaths yearly worldwide (25). The highest mortality is observed in developing countries, such as India, countries of southern Africa, and some South American countries; however, rotavirus infection also constitutes a problem in developed countries, where improved sanitary conditions have not completely eliminated the risk (20). Rotaviruses are 75-nm, round, unenveloped viruses with a three-layer protein capsid. The genome, composed of 11 double-stranded RNA segments, is surrounded by the inner capsid proteins VP1, VP2, and VP3. This complex is surrounded by VP6, which forms the middle layer of the virus capsid. Based on the antigenic characteristics of the VP6 protein, serogroups A to G have been distinguished. The outer layer consists of VP4 (P protein) and VP7 (G protein) (8). Based on VP4 and VP7 neutralizing epitopes, group A rotaviruses are classified using dual serotype P and serotype G designations. According to the nucleotide sequence of VP4 and VP7 genes, P and G genotypes are also distinguished. The most prevalent rotavirus genotypes in humans are G1P[8], G2P[4], G3P[8], G4P[8], and G9P[8] (13). In the last few years, the G12 genotype has apparently spread from Southeast Asia to all continents, similar to the worldwide emergence of G9 strains in the middle 1990s * Corresponding author. Mailing address: Department of Biotechnology and Systems Biology, National Institute of Biology, Vecˇna pot 111, 1000, Ljubljana, Slovenia. Phone: 386 1 4233388. Fax: 386 1 2573847. E-mail: [email protected]. 䌤 Published ahead of print on 4 June 2008. 2547

2548

´ RREZ-AGUIRRE ET AL. GUTIE

J. CLIN. MICROBIOL.

FIG. 1. Design of the qPCR approach. The sequences of the five forward primers, two reverse primers, and degenerate TaqMan MGB probe are shown, followed by the sequence accession numbers and positions to which each primer would anneal without any nucleotide mismatch. The corresponding accession numbers and rotavirus genotype for each of the aligned sequences are shown to the left of the alignment.

it to such samples. In this study, a new RT-qPCR approach for rotavirus detection is presented. The performance of the approach was tested against isolates representative of most prevalent human genotypes and compared with classic diagnostic tools, such as enzyme-linked immunosorbent assay (ELISA) and RT-PCR. The ability to detect a broad range of rotavirus strains involving 17 different genotypes, including human, bovine, and porcine, was also proven. The method shows similar detection abilities whether the viruses are present in buffer, tap water, or environmental water. MATERIALS AND METHODS Sample collection and preparation. Human rotavirus strains were obtained from stool samples from children hospitalized with acute gastroenteritis during the rotavirus seasons of 2003–2004, 2004–2005, and 2005–2006 in the University Medical Centre of Ljubljana and Maribor, Slovenia. All children were ⱕ5 years old, and the samples were collected 1 to 3 days after disease onset. Some of the samples shown (see Tables 1, 2, 3, and 5) were collected from a previous study (26), while other samples (see Table 4) were collected in the work of Steyer et al. (27). Stool samples were diagnosed as rotavirus positive by ELISA (Premier Rotaclone; Meridian Bioscience, Cincinnati, OH) and/or EM (JEM 1200 EXII; Jeol, Tokyo, Japan). Porcine and bovine stool samples were collected mainly from asymptomatic pigs and calves on Slovenian farms at different locations throughout the country. Sampling was performed from January to April and from October to December 2004 and 2005. Genotype characterization of rotavirus isolates was carried out as described below. Stool samples defined as rotavirus negative by means of ELISA or RT-PCR (see below) were also selected for the study, as well as samples positive for other enteric viruses but negative for rotaviruses. The latter consisted of clinical samples from hospitalized patients who tested positive for either astrovirus, norovirus, or sapovirus by ELISA (Oxoid Ltd., United Kingdom) and/or EM. All stool samples were prepared as 10% suspensions in phosphate-buffered saline (0.2 M; pH 7.4). The suspensions were centrifuged for 5 min at 1,600 ⫻ g, and then 250 ␮l of the supernatant was used for total-RNA isolation using Trizol reagent, according to the manufacturer’s instructions (Invitrogen, Carlsbad, CA). Purified RNA was resuspended in 30 ␮l of sterile nuclease-free water (Promega Corporation, Madison, WI) and stored at ⫺80°C. When required, stool samples were diluted to the desired concentrations using healthy stool, Premier sample diluent buffer (Meridian Bioscience Inc., Cincinatti, OH), tap water, or environmental water, and afterwards, 250 ␮l was subjected to total-RNA isolation with Trizol reagent. In some cases, luciferase (luc) RNA (Promega, Madison, WI) was added (2 ng per sample) directly to the Trizol reagent before RNA extraction as an external control for the extraction procedure and to report potential PCRinhibitory effects inherent to the samples. A negative control for the extraction procedure, consisting of buffer only, was always included. Viral-concentration estimation with EM. The rotavirus concentration in stool samples was estimated by virus counting using the latex-negative staining tech-

nique (31). Briefly, three different dilutions of 112-nm beads (1 ⫻ 1010, 1 ⫻ 109, and 1 ⫻ 108 particles/ml) were mixed with three aliquots of the same virus suspension. The original suspension of latex particles was of known concentration (1.29 ⫻ 1012 particles/ml) (Agar Scientific, Stansted, United Kingdom). EM grids were pretreated with Alcian blue to increase adsorption. Negative staining was performed using 1.5% phosphotungstic acid, pH 7. Rotaviruses and latex particles were counted in at least three different locations on the grid. The rotavirus concentration in suspension was estimated by referring to the known concentration of latex particles. Rotavirus detection by seminested RT-PCR. Two microliters of Trizol-isolated total RNA was subjected to RT-PCR using the group A-specific primers Beg9/ End9 for the VP7 gene as described previously (11). To achieve higher sensitivity, seminested PCR was also performed using the RT-PCR product as a template. For the seminested PCR, primers A2 (12) and End9 were used. The products of RT-PCR and nested PCR were loaded on a 2% agarose gel and visualized using ethidium bromide staining on a UV transilluminator. Rotavirus G and P genotyping. For the genotyping of selected rotavirus strains, G and P typing methods were used as described previously (3, 9, 11). Briefly, multiplex nested PCR was carried out separately for the VP7 and VP4 genes, using the amplified products of the first RT-PCR of VP7 (Beg9/End9 primers) and VP4 (Con2/Con3 primers) genes as templates. For G genotyping, G1- to G4-, G8-, and G9-specific primers, and for P genotyping, P[4]-, P[6]-, P[8]-, P[9]-, P[10]-, and P[14]-specific primers were used in the multiplex nested PCRs. The VP4 and VP7 genome segments from isolates that were untypeable by nested RT-PCR were partially sequenced and aligned with sequences from strains of known genotype from public databases for genotype determination. qPCR primer and probe design. Rotavirus VP2 gene sequences from different human isolates representing different genotypes (Fig. 1) were retrieved from the public database of the National Center for Biotechnology Information. Multiplesequence alignments were performed using the ClustalW tool (28) in the Vector NTI-v7 package (Informax Inc., Invitrogen) (Fig. 1). Two reverse and five forward primers, together with a degenerate minor groove binding (MGB) TaqMan probe, were designed in the most conserved area (the 3⬘ end of the VP2 gene) with the help of Primer Express 2.0 software (Applied Biosystems, Foster City, CA). The specific annealing of all primers and probes to the rotavirus VP2 gene was confirmed in silico with a Basic Local Alignment Search Tool (BLAST) search of public databases (http://www.ncbi.nlm.nih.gov/BLAST/BLAST.cgi). Primer and probe sequence information, annealing positions in the corresponding isolate, and sequence accession numbers are detailed in Fig. 1. The MGB probe was labeled with 6-carboxyfluorescein at the 5⬘ end and with MGB probe and a nonfluorescent quencher at the 3⬘ end. luc RNA was used in some cases as an external control for monitoring variations in nucleic acid extraction and the presence of inhibitors. The luc amplicon primers and TaqMan probe sequences, as well as their annealing positions in the luciferase gene, have been described previously (29). RT-qPCR assay. For the RT step, the high-capacity cDNA archive kit (Applied Biosystems, Foster City, CA) was used. The RT master mixture was composed of a deoxynucleoside triphosphate mixture, random hexamers, RT buffer, RNase inhibitor (supplied separately; Applied Biosystems), and MultiScribe reverse transcriptase at concentrations as indicated by the supplier in a final

VOL. 46, 2008

qPCR DETECTION OF ROTAVIRUS

2549

TABLE 1. qPCR performance Linear regression Genotype (strain)

G1P关8兴 (SI-201/06) G2P关4兴 (SI-54/04) G3P关8兴 (SI-233/03) G4P关8兴 (SI-MB7) G9P关8兴 (SI-761/06) G12P关8兴 (SI-264/06)

Concn (particles/ml)

1.7 ⫻ 1011a 6.7 ⫻ 1010 1 ⫻ 1010 9 ⫻ 109 1 ⫻ 1010 3.2 ⫻ 1010

Range of detectionb

Slope

R2

E

⫺3.390 ⫺3.492 ⫺3.233 ⫺3.359 ⫺3.423 ⫺3.432

0.998 0.999 0.998 0.998 0.998 0.999

0.97 0.93 1.03 0.98 0.96 0.95

Neat Neat Neat Neat Neat Neat

to to to to to to

10⫺6 10⫺8 10⫺6 10⫺6 10⫺7 10⫺8

Quantification rangeb

⌬CT luciferase (sample/buffer)

10⫺1 to 10⫺5 Neat to 10⫺7 10⫺1 to 10⫺5 Neat to 10⫺5 Neat to 10⫺5 Neat to 10⫺6

3.3/3.7 3.9/3.7 3.6/3.7 3.6/3.7 3.6/3.7 3.8/3.7

RNA was 10 times diluted before RT-qPCR due to observed inhibition if undiluted RNA was used. ⌬CT luciferase (sample/buffer) ⫽ ⫺14/3.7. Detection and quantification ranges are expressed as the range of cDNA dilutions that can be detected and quantified following the criteria described in Materials and Methods. a b

50-␮l reaction volume. First, 2 ␮l of Trizol (Invitrogen, Carlsbad, CA)-extracted RNA was mixed with the deoxynucleoside triphosphate mixture and random hexamers and subjected to 5 min of incubation at 95°C to allow the denaturation of the rotaviral double-stranded RNA. The mixture was then placed immediately on ice for 2 min, and then the rest of the reactive ingredients were added and RT was performed by incubating the mixture for 10 min at 25°C and 120 min at 37°C. The qPCRs were performed in a 10-␮l final reaction volume, including Taqman universal PCR master mixture (Applied Biosystems), 900 nM of each of the five forward primers, 900 nM of each of the two reverse primers, 250 nM of the MGB TaqMan probe, and 2 ␮l of the corresponding cDNA. For the luc assay, the primer and probe concentrations were 900 nM and 200 nM, respectively. The PCR was performed in 384-well plates (Applied Biosystems). The reactions were run in triplicate on an ABI PRISM 7900HT sequence detection system (Applied Biosystems) using universal cycling conditions (2 min at 50°C and 10 min at 95°C, followed by 45 cycles of 15 s at 95°C and 1 min at 60°C) with 9,600 emulation modes. The threshold cycle (CT) for each individual amplification was obtained using SDS 2.2 software (Applied Biosystems). For all calculations, the baseline was set automatically and the fluorescence threshold manually. Nontemplate controls were used to monitor for potential contamination within the qPCR reagents. RT-qPCR performance. Parameters such as amplification efficiency, range of detection, and range of quantification were determined for evaluation of the performance characteristics of the developed method on six isolates representing each of the most prevalent human rotavirus genotypes. For this purpose, total RNA was Trizol isolated from stool samples infected with a single genotype and with a known preestimated viral load. A known concentration (2 ng per sample) of luc RNA was added directly to the Trizol before the RNA extraction. The RNA was transcribed to cDNA, and nine 10-fold serial dilutions were subjected in triplicate to VP2 amplicon qPCR detection. At the same time, neat and 10-times-diluted cDNAs were subjected to luc amplicon detection. The CT value of the luc in undiluted cDNA and the ⌬CT between neat and 10-times-diluted cDNA samples were evaluated to search for potential variations in the RNA extraction procedure and for the presence of reaction inhibitors, respectively. With the obtained VP2 CT values, standard calibration curves were built by plotting the CT values versus the logarithm of the estimated viral-RNA concentration corresponding to each cDNA dilution. The equations of the linear regression and the R2 value were then obtained for each curve. From the slope of the regression curves, the amplification efficiency (E) was calculated according to the following equation: E ⫽ 10(⫺1/slope) ⫺ 1, where a value of 1 corresponds to 100% efficiency (10, 21). The viral-RNA concentration associated with each cDNA dilution was recalculated based on the obtained regression equations and was used for calculating the coefficient of variance within each triplicate. cDNA dilutions whose triplicate CT values gave a coefficient of variance lower than 30% were assumed to be within the range of quantification (4, 5). In regard to the range of detection, cDNA dilutions that gave a positive CT value in at least two of the triplicates were considered to be within the range of detection. cDNA dilutions giving a positive CT value in only one of the triplicates, as well as those for which CT values were equal to or above 40, were always considered negative.

RESULTS Design of the qPCR system. For the design of the RT-qPCR system, the rotaviral VP2 gene was used as a target. This segment of the rotavirus genome is not involved in the G-P classification. The G types G1, G2, G3, G8, and G12 were the

only human genotypes for which the whole VP2 gene sequence was available in public databases. We included six sequences in the alignment, representing the above-mentioned genotypes and corresponding to strains TB-Chen (G2P[4]), B4106 (G3P[14]), DRC86 (G8P[6]), RV176 (G12P[6]), Wa (G1P[8]), and Ku (G1P[8]) (Fig. 1). The whole VP2 gene shared 71.3% identity among the aligned sequences, while the selected qPCR target region shared 78.5%. Aiming to overcome the high level of sequence variability, we designed multiple forward and reverse primers and a degenerate MGB TaqMan probe. In designing these primers, we tried to minimize the potential negative effect due to the formation of primer dimers and to primer secondary structure by using Primer express v. 2.0 (Applied Biosystems). The in silico analysis predicted that all aligned sequences could be amplified and detected by mixing in the qPCR five different forward and two different reverse primers, together with an MGB probe degenerate at three positions (Fig. 1). RT-qPCR performance on the most prevalent human genotypes. The next step was to experimentally test the performance of our approach. For this purpose, we used six human stool samples, each of them infected with one of the most prevalent human genotypes, namely, G1P[8], G2P[4], G3P[8], G4P[8], G9P[8], and G12P[8]. These samples were genotyped and characterized in a previous study (26). The concentration of rotavirus in the samples was estimated by particle counting under the electron microscope and using latex beads of known size and concentration as standards. All the samples presented rotaviral loads within the range of 1 ⫻ 109 to 1 ⫻ 1011 particles/ml (Table 1). Total RNA was extracted from each sample and reverse transcribed to cDNA, and 10-fold serial dilutions (neat to 10⫺9) of the cDNA were subjected to qPCR in triplicate. luc RNA was added to the Trizol prior to the RNA extraction as an external control to monitor for any potential inhibitory effect inherent to the samples. The increment in the CT value, ⌬CT, from neat to 10-times-diluted cDNA was calculated for the luciferase (⌬CTluc) added in a buffer sample and compared to the ⌬CTluc observed in the stool samples. ⌬CTluc values in all stool samples were similar to the value of 3.9 observed in buffer (Table 1), indicating lack of inhibition. The only exception was sample G1P[8], where luciferase detection was strongly inhibited (a CT of 38.3 in undiluted cDNA in comparison to a CT of 24.75 in 10-times-diluted cDNA; ⌬CTluc ⫽ ⫺13.6), probably due to the large amount of RNA present in this sample, which was, by far, the one presenting

2550

´ RREZ-AGUIRRE ET AL. GUTIE

the highest viral load (1.7 ⫻ 1011 particles/ml). By diluting 10 times the RNA extracted from this sample before reverse transcription, the inhibition was almost abolished (a ⌬CTluc of 3.3, in comparison to a ⌬CTluc of 3.9 in buffer). All of the samples showed similar CT values for luciferase in the undiluted cDNA (CT ⬇ 17; data not shown), indicating that the RNA extractions were similarly efficient in all cases. The addition of luciferase RNA prior to total-RNA extraction proved to be a suitable control both for testing for the presence of inhibitors and for evaluating variations in the extraction procedure. In Table 1, we show the qPCR amplification parameters for each rotavirus genotype analyzed. All six tested strains were successfully amplified and detected in real time with the developed system. The CT values obtained at each cDNA dilution were plotted against the logarithm of the estimated concentration, and linear regressions were obtained. The R2 for all the regressions was above 0.998. From the slope of the regression’s equation, the efficiency of the amplification reaction was calculated, which was in all cases between 0.93 and 1.03 (a value of 1.00 means 100% efficiency). The range of detection varied depending on the strain analyzed, reaching limits of detection that ranged from 10⫺6 cDNA dilution in the cases of G1P[8], G3P[8], and G4P[8] to 10⫺8 cDNA dilution in the cases of G2P[4] and G12P[8] (Table 1). The range of quantification also varied with the sample’s genotype composition; it was lowest (G1P[8] and G3P[8]) from 10⫺1 to 10⫺5 cDNA dilution, while in the case of G2P[4], this range increased from neat to 10⫺7 cDNA dilution (Table 1). The variations in the limits of detection and quantification among genotypes were most probably related to differences at the sequence level within the VP2 amplicon. Limit of detection in stools: viral, RNA, and cDNA dilutions. To test the ability of the method to detect low concentrations of rotaviruses in stools, we performed a 10-fold serial dilution of a rotavirus sample (4.17 ⫻ 109 particles/ml) in healthy stool and then subjected each dilution to total RNA isolation and RT-qPCR detection. We selected a G2P[4] rotavirus, the one most abundantly detected in stools in Slovenia in the 2006– 2007 season. The method could detect rotavirus up to a 10⫺7 dilution, meaning a limit of detection of 4.17 ⫻ 102 particles/ml (Table 2). This limit of detection, as stated above, may vary depending on the particular strain present in the stool. To test the equivalency between diluting virus particles and diluting viral RNA or viral cDNA, we also performed 10-fold serial dilutions of the most concentrated RNA sample and cDNA sample in RNase free H2O and in sterile H2O, respectively. All the RNA dilutions were subjected to RT-qPCR detection, while all cDNA dilutions were directly subjected to qPCR detection. As shown in Table 2, the results obtained when diluting viral particles, RNA, and cDNA were equivalent. These results also indicate a nice reproducibility of the method, both at high concentrations (10⫺1 dilution) and at low concentrations (10⫺6 dilution). In the case of the cDNA dilutions, the last dilution was not detected due to the proximity of the limit of detection (CT ⬎ 35), where interassay reproducibility decreases. Comparison with other diagnostic methods. Next, we compared the ability of the method to detect rotavirus RNA in stools with those of classic routine diagnostic tools, such as ELISA and seminested RT-PCR. For this purpose, two differ-

J. CLIN. MICROBIOL. TABLE 2. Limits of detection in stool: viral, RNA, and cDNA dilutions Limit of detection (mean CT) Sample dilution

Virus dilution (healthy stool)a

RNA dilution (RNase-free H2O)b

cDNA dilution (sterile H2O)c

10⫺1 10⫺2 10⫺3 10⫺4 10⫺5 10⫺6 10⫺7 10⫺8 Backgroundd

⫹17.99 ⫹21.79 ⫹25.07 ⫹28.69 ⫹32.14 ⫹35.76 ⫹38.49 Undetermined Undetermined

⫹18.16 ⫹21.71 ⫹25.40 ⫹28.92 ⫹32.17 ⫹36.51 ⫹38.41 Undetermined Undetermined

⫹18.79 ⫹22.29 ⫹25.71 ⫹29.31 ⫹32.83 ⫹36.22 Undetermined Undetermined Undetermined

a Virus sample (rotavirus G2P关4兴; 4.17 ⫻ 109 part/ml) was 10-fold serially diluted in healthy stool. Each dilution was subjected to RNA isolation and RT-qPCR. b After RNA isolation, the most concentrated RNA sample was 10-fold serially diluted in RNase-free H2O. Each dilution was subjected to RT-qPCR. c After the RT step, the most concentrated cDNA sample was 10-fold serially diluted in sterile H2O. Each dilution was subjected to qPCR. d As a background, healthy stool, RNase-free H2O, and sterile H2O, respectively, were used.

ent stool samples were prepared, one of them containing a single G2P[4] rotavirus genotype and a second one composed of a mixture of that genotype and G1P[8]. G1P[8] was the most frequently detected genotype in Slovenia during the 2005–2006 season (26), while G2P[4] was the most prevalent during 2006– 2007. The viral loads of both stool samples were close to 1 ⫻ 1010 particles/ml as estimated by EM counting. We performed 10-fold serial dilutions of both samples in sample diluent buffer. Each dilution was subjected to ELISA detection, and following that, total RNA was extracted and subjected to RTqPCR detection and to seminested RT-PCR (Table 3). The RT-qPCR detected 1 ⫻ 103 and 1 ⫻ 104 particles/ml of G2P[4] and the mixture G1P[8]-G2P[4], respectively, improving the ELISA detection by 6 and 5 orders of magnitude in each case (Table 3). Detection with both RT-qPCR and seminested RTPCR varied depending on the sample, again most probably due to sequence variations in the corresponding amplicon in each case. Seminested RT-PCR detected the sample composed of G1P[8]-G2P[4] 2 orders of magnitude better (1 ⫻ 105 particles/ ml) than the G2P[4] sample (1 ⫻ 107 particles/ml), while RTqPCR detected the G2P[4] sample (1 ⫻ 103 particles/ml) 1 order of magnitude better than the mixture (1 ⫻ 104 particles/ ml) (Table 3). RT-qPCR detection worked better in both samples than seminested RT-PCR by 4 and 1 orders of magnitude in stools composed of G2P[4] and G1P[8]-G2P[4], respectively (Table 3). Survey of human and animal clinical samples. Next, we decided to test the method on a broad range of different rotavirus-positive clinical samples collected, analyzed, genotyped, and stored at the Institute of Microbiology of the Faculty of Medicine of Ljubljana, Slovenia, during the years 2004 to 2006. The samples consisted of total RNA isolated from stools infected with different genotypes of human (121 samples), bovine (6 samples), and porcine (12 samples) origin (Table 4). Additionally, we analyzed RNAs coming from 10 stool samples classified as rotavirus negative by means of ELISA and/or RT-PCR (Table 4). Among samples of human

VOL. 46, 2008

qPCR DETECTION OF ROTAVIRUS

TABLE 3. Comparison of the developed RT-qPCR system with seminested RT-PCR and ELISA RT-qPCR (mean CT)

Seminested RT-PCRa

ELISAa

G2P关4兴 1.4 ⫻ 1010 1.4 ⫻ 109 1.4 ⫻ 108 1.4 ⫻ 107 1.4 ⫻ 106 1.4 ⫻ 105 1.4 ⫻ 104 1.4 ⫻ 103 1.4 ⫻ 102 1.4 ⫻ 101 Buffer

⫹11.8 ⫹16.5 ⫹21.6 ⫹25.6 ⫹29.4 ⫹31.9 ⫹35.5 ⫹37.2 Undetermined Undetermined Undetermined

⫹ ⫹ ⫹ ⫹ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺

⫹ ⫹ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺

G1P关8兴 ⫹ G2P关4兴 1.3 ⫻ 1010 1.3 ⫻ 109 1.3 ⫻ 108 1.3 ⫻ 107 1.3 ⫻ 106 1.3 ⫻ 105 1.3 ⫻ 104 1.3 ⫻ 103 1.3 ⫻ 102 1.3 ⫻ 101 Buffer

⫹13.8 ⫹17.9 ⫹22.9 ⫹27.2 ⫹30.7 ⫹34.6 ⫹37.1 Undetermined Undetermined Undetermined Undetermined

⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫺ ⫺ ⫺ ⫺ ⫺

⫹ ⫹ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺

Concn (parts/ml)

a ELISA results were considered positive (⫹) when the optical density value was above 0.15 (according to kit manufacturer instructions). Seminested RTPCR results were considered positive when the expected amplification product (257 bp) was visible on the agarose gel. ⫺, negative result.

origin, we analyzed the most prevalent genotypes (G1P[8], 63 samples; G2P[4], 20 samples; and G9P[8], 29 samples), but also more atypical ones (G4P[8], 7 samples; G8P[8], 1 sample; and G12P[8], 1 sample) (Table 4). All 121 human samples were detected by RT-qPCR; 56% of the samples had CT values between 10 and 20 (including the most atypical genotypes), 40% had CT values between 20 and 30, and just 4% had CT values higher than 30 (Table 4). Strains of animal origin were also analyzed (Table 4). Such strains are expected to be present in environmental water samples; therefore, it was interesting to confirm if the assay could detect them, as well. Six bovine and 12 porcine samples were tested, including four and seven different genotypes, respectively. Again, all samples were successfully detected by the newly developed RT-qPCR method. Among bovine samples, five out of six samples were detected at CT values between 10 and 30, while one sample, corresponding to an undetermined genotype, was detected at CT values higher than 30. For the porcine samples, half of them, corresponding to the most prevalent porcine genotype, G5P[7], were detected within a CT value range of 10 to 30, while the other six analyzed genotypes were detected at CT values higher than 30. In order to prevent potential inhibition of the qPCR, all cDNAs shown in Table 4 were 10 times diluted before being subjected to the reaction mixture. When we analyzed the 10 samples that were classified as negative using classical detection methods, we found that one of the samples was positive by RT-qPCR with a CT of 38.2 with 10-times-diluted cDNA and a CT of 36.2 with undiluted cDNA.

2551

To discard a false-positive result, we repeated the reaction in the absence of the MGB probe using Sybr green chemistry, and we performed a melting-curve analysis. Substantial differences between the melting temperature (Tm) of the sample and the one associated with the designed VP2 amplicon would indicate a false positive. The obtained Tm (74.6°C) was very similar to the Tm obtained with a known rotavirus, G1P[8], cDNA (74.5°C), thus confirming a rotavirus-positive result. As already shown in Table 3, this confirms the higher sensitivity of the developed qPCR approach in comparison with ELISA and RT-PCR, which are the routinely used diagnostic tools. Specificity of the assay. In the course of the qPCR design, all primers and probes were subjected to BLAST analysis, and significant hits restricted uniquely to the rotavirus VP2 gene were obtained. In order to experimentally confirm the rotavirus specificity of the developed approach, we decided to additionally test 10 samples that were positive for other enteric viruses, including norovirus GG.I (three samples), norovirus GG.II (three samples), astrovirus (three samples), and sapovirus (one sample) (Table 4). All of them gave negative results with RT-qPCR, except the three norovirus GG.II samples, which gave positive results with CT values between 35 and 38. It is known that low titers of rotavirus can be present together with norovirus in the form of double infections (2), but in order to discard the possibility of cross-reaction with norovirus GG.II, we subjected the three RNA samples to seminested RT-PCR rotavirus detection, and additionally, we performed melting-curve analysis to check the specificities of the qPCR products, as explained above. Melting-curve analysis resulted in a Tm of 74.7°C for all three samples, indicating rotavirusspecific detection. Seminested RT-PCR confirmed the rotavirus-positive results in two of the three suspicious samples, showing a weak band in each case. The third sample was negative by seminested RT-PCR, probably due to the low rotavirus titer, which was also shown by the high CT values obtained in the more sensitive RT-qPCR assay. The lack of cross-reactivity with norovirus was further confirmed by applying RNA, generated by in vitro transcription of a GG.II cDNA, to the rotavirus-specific RT-qPCR, resulting in negative results. Detection in environmental water samples. Apart from routine diagnostic testing for rotavirus in clinical samples, we also intend to apply the developed method to the surveillance of environmental waters for the presence of rotavirus. To test if the developed method is able to detect the presence of rotaviruses in such samples, we used a rotavirus sample (2.6 ⫻ 1010 particles/ml) consisting of a mixture of genotypes and performed 10-fold serial dilutions in buffer, potable tap water from the lavatories of the National Institute of Biology, and water from a stream that flows close to the same facilities. An unknown genotype composition was generated, as that best resembled what might be present in such samples after, for example, the occurrence of an earthquake or flood. All the tested concentrations were detected similarly, whether the virus was present in sample diluent buffer, stream water, or tap water (Table 5), indicating that the method could be applied to such matrices. This result again addresses the reproducibility of the assay regardless of the matrix in which viruses are present. In the samples of 2.6 ⫻ 104 particles/ml, we obtained

2552

´ RREZ-AGUIRRE ET AL. GUTIE

J. CLIN. MICROBIOL.

TABLE 4. Validation of the developed system through analysis of 159 clinical samples from years 2004 to 2006 in Slovenia No. of samples

No. RT-qPCR positive

No. 10 ⱕ CT ⱕ 20

No. 20 ⬍ CT ⱕ 30

No. 30 ⬍ CT ⬍ LODb

Human samples G1P关8兴 G2P关4兴 G4P关8兴 G8P关8兴 G9P关8兴 G12P关8兴 Total

63 20 7 1 29 1 121

63 20 7 1 29 1 121 (100%)

24 19 7 0 17 1 68 (56%)

35 1 0 1 11 0 48 (40%)

4 0 0 0 1 0 5 (4%)

Bovine samples G6P关1兴 G6P关5兴 G6P关11兴 NDa Total

2 2 1 1 6

2 2 1 1 6 (100%)

0 2 1 0 3 (50%)

2 0 0 0 2 (33%)

0 0 0 1 1 (17%)

Porcine samples G2P关13兴 G3P关6兴 G4P关22兴 G5P关7兴 G5P关13兴 G9P关13兴 G11P关13兴 Total

1 1 1 6 1 1 1 12

1 1 1 6 1 1 1 12 (100%)

0 0 0 4 0 0 0 4 (33%)

0 0 0 2 0 0 0 2 (17%)

1 1 1 0 1 1 1 6 (50%)

Rotavirus-negative samples ND Total

10 10

1 1

0 0

0 0

1 1

Samples positive for other enteric viruses Norovirus GI Norovirus GII Astrovirus Sapovirus Total

3 3 3 1 10

0 3 0 0 3

0 0 0 0 0

0 0 0 0 0

0 3 0 0 3

Genotype

a b

ND, not determined. LOD, limit of detection.

a negative result in the case of the tap water, in contrast to the other two diluents, but as stated above, this was most probably due to the proximity to the limit of detection (CT ⬎ 35), where stochastic effects gain more importance. Therefore, for the

TABLE 5. Rotavirus detection in artificially inoculated tap and environmental waters Rotavirus mixturea in diluent: Concn (parts/ml)

2.6 ⫻ 109 2.6 ⫻ 108 2.6 ⫻ 107 2.6 ⫻ 106 2.6 ⫻ 105 2.6 ⫻ 104 2.6 ⫻ 103 No virus

Buffer (mean CT)

Tap waterb (mean CT)

Environmental waterc (mean CT)

⫹21.06 ⫹26.25 ⫹29.25 ⫹32.86 ⫹35.9 ⫹38.27 Undetermined Undetermined

⫹21.15 ⫹25.67 ⫹30.01 ⫹33.68 ⫹36.92 Undetermined Undetermined Undetermined

⫹21.99 ⫹26.73 ⫹30.03 ⫹33.54 ⫹35.85 ⫹38.65 Undetermined Undetermined

a Rotavirus sample obtained by randomly mixing different rotavirus genotypes archived at the Institute of Microbiology and Immunology, Ljubljana, Slovenia. b Tap water collected from the lavatories of the National Institute of Biology, Ljubljana, Slovenia. c Water collected from a stream close to the National Institute of Biology, Ljubljana, Slovenia.

particular case of this mixture of genotypes, the limit of detection in environmental waters was in the range of 104 to 105 particles/ml. Lower limits of detection are expected in detecting a single genotype, such as G2P[4].

DISCUSSION The main objective of this study was to design a simple, fast, and sensitive RT-qPCR-based method for the detection of a broad range of rotavirus genotypes to be applied both to routine monitoring of stool samples and to the detection of rotaviruses present in environmental samples, such as tap water, recreational waters, or river/lake waters. Rotavirus genotypes are increasing in number in both humans and animals, and environmental water samples may represent a reservoir for a broad range of rotavirus genotypes (6, 30). Moreover, rotaviruses of animal origin, with a demonstrated potential for interspecies transmission, constitute an added risk for the infection of humans (6, 16). The few TaqMan RT-qPCR approaches published did not succeed in detecting more than one (G1) (15) or three (G1, G2, and G4) (19) human rotavirus types. If we focus on rotavirus of animal origin, the situation is worse, as we were not able to find any data on RT-qPCR

VOL. 46, 2008

detection of such genotypes. In the first of the above-mentioned studies, Logan et al. found surprising the detection of just a single type (G1) in the survey they performed in Ireland with a VP6 gene-directed Taqman RT-qPCR assay. They analyzed a total of 139 stool samples from patients exhibiting typical gastroenteritis symptoms, and 20 of them were rotavirus positive by the developed RT-qPCR assay. The reason for finding just the G1 type could be the small number of samples (nine) that were further genotyped, but at the same time, a systematic analysis of the strain/genotype coverage of the assay using known genotype-positive controls was missing. Due to the high level of variability among rotavirus isolates, it could be that some of the genotypes escaped detection. To our knowledge, this is the first study in which a TaqMan RT-qPCR detection assay has been systematically tested against a broad range of both human and animal rotavirus strains. Selecting an amplicon target (VP2) that is independent of rotavirus genome segments involved in G-P classification (VP4 and VP7) enabled detection of rotavirus strains independently of their G-P genotype compositions. The potential of the approach for the detection of the broadest range of rotavirus strains is further enhanced by the presence in the PCR of a mixture of five forward and two reverse primers in combination with a degenerate TaqMan MGB probe. The method successfully detected the presence of rotavirus in 139 isolates (Table 4), involving 17 different G-P types divided into 7 of human origin (G1P[8], G2P[4], G3P[8], G4P[8], G8P[8], G9P[8], and G12P[8]), 3 of bovine origin (G6P[1], G6P[1] G6P[5], and G6P[11]), and 7 of porcine origin (G2P[13], G3P[6], G4P[22], G5P[7], G5P[13], G9P[13], and G11P[13]). The flexibility of the developed approach did not compromise its sensitivity. Of the 139 stool samples analyzed, including human, bovine, and porcine, only 12 were detected at CT values higher than 30 (Table 4). The mixture of primers and the degenerate probe did not affect the efficiency of the amplification, since values close to 1 were obtained when the qPCR parameters for six isolates representing the most prevalent human genotypes were analyzed (Table 1). The limit of detection for such isolates was in the range of 10⫺6 to 10⫺8 cDNA dilutions, depending on the strain (Table 1), which, assuming a 100% yield of the RNA extraction and RT steps and taking into account that the viral load of the tested samples was close to 1 ⫻ 1010 particles/ml, results in an ability to detect approximately 1 ⫻ 102 to 1 ⫻ 104 particles/ml in the initial sample. This was further confirmed by diluting a G2P[4] rotavirus in healthy stool, reaching a limit of detection of 4.17 ⫻ 102 particles/ml (Table 2). We also showed that the tested isolates could be readily quantified using the obtained standard curves with limits of quantification ranging from 10⫺5 to 10⫺7 cDNA dilution, depending on the isolate (Table 1). However, if an accurate absolute quantification is desired, further and more thorough controls addressing the losses in RNA extraction and RT steps, as described by Costafreda et al. (7), are required. The differences observed in the CT values obtained when the samples shown in Table 4 were detected could be related to a variety of factors, i.e., different rotaviral loads in the original samples or different integrities of the RNAs, which in some cases had been stored since 2004. However, these differences may also arise from sequence variations within the amplicon

qPCR DETECTION OF ROTAVIRUS

2553

site from sample to sample. As shown in Table 1, although the viral loads and amplification efficiencies were similar in most of the samples, the limits of detection and quantification varied depending on the isolate. A similar effect was previously noticed in different qPCR applications, such as detection of an amplicon targeted to a proteinase gene (24) or quantification of genetically modified organisms present in foods (18). Similarly to our case, small sequence level variations in the amplicon region caused a shift in the CT value with no effect on the amplification efficiency. The developed assay represents a powerful diagnostic tool for routine detection of rotavirus in stool samples (Tables 2 and 4). Detection was improved over that with ELISA and seminested RT-PCR (Table 3) and enabled the finding of rotavirus in samples in which those techniques failed to detect them (Table 4). At the same time, RT-qPCR minimizes the always problematic cross-contamination, as real-time detection avoids the handling of PCR products. The use of 384-well plates allowed the simultaneous analysis of 120 cDNA samples in triplicate, considerably reducing the time for analysis in comparison with RT-PCR. Furthermore, the flexibility of the assay ensures the detection of typical and atypical genotypes of human and animal origin with a single experimental design. This flexibility can be exploited in the detection of low concentrations of rotaviruses present in environmental waters, where different rotavirus genotypes are expected depending on the contamination source. We already confirmed that the method can detect rotaviruses spiked in such samples (Table 5), and the broad genotype coverage would decrease the probability of obtaining false negatives. The combination of such a sensitive and, at the same time, flexible detection system with a proper concentration step will allow the efficient survey of water samples in search of minimal concentrations of rotavirus, thus helping to localize the contamination sources and therefore reducing the risk for the population. ACKNOWLEDGMENTS This work was supported by the Slovenian Research Agency and the Slovenian Ministry of Defense through grant M1-0145 in the call for targeted research projects Research for Security and Peace, 2006– 2009, CRP-MIR. REFERENCES 1. Abad, F. X., R. M. Pinto ´, and A. Bosch. 1994. Survival of enteric viruses on environmental fomites. Appl. Environ. Microbiol. 60:3704–3710. 2. Amar, C. F. L., C. L. East, J. Gray, M. Iturriza-Go ´mara, E. A. Maclure, and J. McLauchlin. 2007. Detection by PCR of eight groups of enteric pathogens in 4,627 faecal samples: re-examination of the English case-control Infectious Intestinal Disease Study (1993–1996). Eur. J. Microbiol. Infect. Dis. 26:311–323. 3. Arista, S., E. Vizzi, C. Alaimo, D. Palermo, and A. Cascio. 1999. Identification of human rotavirus strains with the P[14] genotype by PCR. J. Clin. Microbiol. 37:2706–2708. ˇ trancar, 4. Boben, J., P. Kramberger, N. Petrovicˇ, K. Cankar, M. Peterka, A. S and M. Ravnikar. 2007. Detection and quantification of tomato mosaic virus in irrigation waters. Eur. J. Plant Pathol. 118:59–71. 5. Burns, M. J., H. Valdivia, and N. Harris. 2004. Analysis and interpretation of data from real-time PCR trace detection methods using quantitation of GM soya as a model system. Anal. Bioanal. Chem. 378:1616–1623. 6. Cook, N., J. Bridger, K. Kendall, M. Iturriza-Go ´mara, L. El-Attar, and J. Gray. 2004. The zoonotic potential of rotavirus. J. Infect. 48:289–302. 7. Costafreda, M. I., A. Bosch, and R. M. Pinto´. 2006. Development, evaluation, and standardization of a real-time TaqMan reverse transcription-PCR assay for quantification of hepatitis A virus in clinical and shellfish samples. Appl. Environ. Microbiol. 72:3846–3855. 8. Estes, M. K. 2001. Rotaviruses and their replication, p. 1747–1785. In D. M. Knipe, P. M. Howley, D. E. Griffin, R. A. Lamb, M. A. Martin, B. Roizman

2554

9.

10. 11.

12.

13.

14.

15.

16.

17.

18.

19.

´ RREZ-AGUIRRE ET AL. GUTIE

and S. E. Strais (ed.), Fields virology, 4th ed. Lippincot William and Wilkins, Philadelphia, PA. Gentsch, J. R., R. I. Glass, P. Woods, V. Gouvea, M. Gorziglia, B. K. Das, and M. K. Bhan. 1992. Identification of group A rotavirus gene 4 types by polymerase chain reaction. J. Clin. Microbiol. 30:1365–1373. Ginzinger, D. G. 2002. Gene quantification using realtime quantitative PCR: An emerging technology hits the mainstream. Exp. Hematol. 30:503–512. Gouvea, G., R. I. Glass, P. Woods, K. Taniguchi, H. F. Clark, B. Forrester, and Y. Z. Fang. 1990. Polymerase chain reaction amplification and typing of rotavirus nucleic acid from stool specimens. J. Clin. Microbiol. 28:276–282. Gouvea, V. 1995. PCR detection of rotaviruses, p. 383–387. In D. L. Wiedrauk and D. H. Farkas (ed.), Molecular methods for virus detection. Academic Press, New York, NY. Iturriza-Go ´mara, M., U. Desselberger, and J. Gray. 2003. Molecular epidemiology of rotaviruses: genetic mechanisms associated with diversity, p. 317– 344. In U. Desselberger and J. Gray (ed.), Viral gastroenteritis. Elsevier, Amsterdam, The Netherlands. Kang, G., M. Iturriza-Go ´mara, J. G. Wheeler, P. Crystal, B. Monica, S. Ramani, B. Primrose, P. D. Moses, and C. I. Gallimore. 2004. Quantitation of group A rotavirus by real-time reverse-transcription-polymerase chain reaction: correlation with clinical severity in children in south India. J. Med. Virol. 73:118–122. Logan, C., J. J. O’Leary, and N. O’Sullivan. 2006. Real-time reverse transcription-PCR for detection of rotavirus and adenovirus as causative agents of acute viral gastroenteritis in children. J. Clin. Microbiol. 44:3189–3195. Matthijnssens, J., M. Rahman, V. Martella, Y. Xuelei, S. De Vos, K. De Leener, M. Ciarlet, C. Buonavoglia, and M. Van Rast. 2006. Full genomic analysis of human rotavirus strain B4106 and lapine rotavirus strain 30/96 provides evidence for interspecies transmission. J. Virol. 80:3801–3810. Min, B.-S., Y.-J. Noh, J.-H. Shin, S.-Y. Baek, K.-I. Min, S.-R. Ryu, B.-G. Kim, M.-K. Park, S.-E. Choi, E.-H. Yang, S.-N. Park, S.-J. Hur, and B.-Y. Ahn. 2006. Assessment of the quantitative real-time polymerase chain reaction using a cDNA standard for human group A rotavirus. J. Virol. Methods 137:280–286. Moor, D. 2005. Reliability of the maize and soybean endogenous reference genes used in GMO analysis of food. Ph. D. thesis. University of Bern, Bern, Switzerland. Pang, X. L., B. Lee, N. Boroumand, B. Leblanc, J. K. Preiksaitis, and C. C. Yu Ip. 2004. Increased detection of rotavirus using a real time reverse

J. CLIN. MICROBIOL.

20.

21. 22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

transcription-polymerase chain reaction (RT-PCR) assay in stool specimens from children with diarrhea. J. Med. Virol. 72:496–501. Parashar, U. D., E. G. Hummelman, J. S. Bresee, M. A. Miller, and R. I. Glass. 2003. Global illness and deaths caused by rotavirus disease in children. Emerg. Infect. Dis. 9:565–571. Pfaffl, M. W. 2001. A new mathematical model for relative quantification in real-time PCR. Nucleic Acids Res. 29:2002–2007. Rahman, M., J. Matthijnssens, X. Yang, T. Delbeke, I. Arijs, K. Taniguchi, M. Iturriza-Go ´mara, N. Iftekharuddin, T. Azim, and M. Van Rast. 2007. Evolutionary history and global spread of the emerging G12 human rotaviruses. J. Virol. 81:2382–2390. Schwarz, B.-A., R. Benge, T. W. Vahlenkamp, R. Johne, and H. Mu ¨ller. 2002. Detection and quantitation of group A rotaviruses by competitive and realtime reverse transcription-polymerase chain reaction. J. Virol. Methods 105: 277–285. Slapar, N. 2005. Molecular aspect of Colorado potato beetle (Leptinotarsa decemlineata Say) adaptation to plant defense response. Ph.D. thesis. University of Ljubljana, Ljubljana, Slovenia. Soriano-Gabarro ´, M., J. Mrukowicz, T. Vesikari, and T. Verstraeten. 2006. Burden of rotavirus disease in European Union countries. Pediatr. Infect. Dis. J. 25:7–11. Steyer, A., M. Poljsˇak-Prijatelj, T. Luzˇnik-Bufon, N. Marcˇun-Varda, and J. Marin. 2007. Rotavirus genotypes in Slovenia: unexpected detection of G8P[8] and G12P[8] genotypes. J. Med. Virol. 79:626–632. Steyer, A., M. Poljsˇak-Prijatelj, D. Barlicˇ-Maganja, and J. Marin. 2008. Human, porcine and bovine rotaviruses in Slovenia: evidence of interspecies transmission and genome reassortment. J. Gen. Virol. 89:1690–1698. Thompson, J. D., D. G. Higgins, and T. J. Gibson. 1994. CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res. 22:4673–4680. ˇ el. 2004. A Toplak, N., V. Okrsˇlar, D. Stanicˇ-Racman, K. Gruden, and J. Z high throughput method for quantifying transgene expression in transformed plants with real-time PCR analysis. Plant Mol. Biol. Rep. 22:237–250. van Zyl, W. B., N. A. Page, W. O. Grabow, A. D. Steele, and M. B. Taylor. 2006. Molecular epidemiology of group A rotaviruses in water sources and selected raw vegetables in southern Africa. Appl. Environ. Microbiol. 72: 4554–4560. Zheng, Y. Z., R. Webb, P. F. Greenfield, and S. Reid. 1996. Improved method for counting virus and virus like particles. J. Virol. Methods 62:153–159.